Singapore AI News & Daily Briefing

Singapore is setting up a dedicated institute to drive AI adoption in the financial sector, aiming to keep the city-state competitive as banks and fintech firms globally race to integrate AI tools. The initiative signals stronger government-backed push to blend AI capabilities with Singapore's core financial services industry, which could shape how ordinary Singaporeans bank, invest, and access financial services in the years ahead.

Alibaba.com is rolling out an agentic AI toolset aimed at Malaysian small businesses struggling with talent shortages, helping them automate cross-border selling tasks like inventory and logistics. The promise is 24/7 automated store management, but experts warn that removing human oversight too quickly could introduce financial risks. Malaysian SMEs make up over 96% of local businesses, so success here could be a major test case for AI adoption across Southeast Asia.

Singapore is urging young people to keep thinking for themselves as AI takes on more tasks. The message is clear: do not rely entirely on AI for decisions, especially in work and daily life. The government wants the next generation to use AI as a tool while maintaining critical thinking and human judgment.

IBM has announced what it claims is the world's first sub-1 nanometer chip design using a new transistor architecture called 'nanostack,' packing nearly 100 billion transistors onto a single chip. However, IBM has no manufacturing partner or actual chip yet — it's a design milestone without a factory to build it. This highlights how far chip design has come, even as actually manufacturing such tiny chips remains a massive engineering hurdle.

A Taiwanese tech firm has unveiled its first humanoid robot, joining a growing wave of manufacturers turning AI into physical machines for factories and warehouses. TM Technology, originally an integrated circuit design company, says the robot marks its entry into the Physical AI space — where AI brains get robot bodies. This signals a shift in Taiwan's manufacturing muscle moving beyond just making chips to actually building AI-powered workers.

Huawei has launched 'GigaUplink,' a mobile network technology designed to solve a hidden crisis: AI features on phones generate massive amounts of data that needs uploading, but current networks struggle to handle it. GigaUplink boosts upload speeds fivefold using smarter antenna setups and spectrum management. As AI features in phones become standard, this bottleneck will affect every smartphone user — whether they're video-calling, live-streaming, or running on-device AI tasks.

OpenAI has previewed GPT-5.6 Sol, its next-generation AI model, on the company's official blog. The preview marks an upcoming release in the GPT series, though technical details about capabilities were not fully available in the announcement. The GPT-5 series represents the next major step in OpenAI's language model development.

Apple has fixed a serious security flaw in its Beats Studio Buds earbuds that could have allowed nearby hackers to secretly listen through your phone's microphone. The vulnerability, rated 8.8 out of 10 in severity, affected the Bluetooth chip firmware made by Airoha Systems. Apple released a patch (Firmware Update 1B211) that installs automatically. Other brands like Bose, JBL, and Jabra were also affected. While no real-world attacks have been reported, the complexity of such exploits means users should ensure their earbuds are updated.

IEEE has launched a five-course online training program called 'Large Language Models Demystified' to help technical professionals understand and build with LLMs. The program covers transformer architecture, PyTorch training, and deployment. The LLM technology market is growing about 33% yearly through 2030. Participants earn professional development credits and a digital badge upon completion. The courses are available through the IEEE Learning Network.

Four Singapore companies and three individuals face fraud and money laundering charges over allegedly illegal exports of servers possibly containing restricted Nvidia AI chips. The Aperia Group and Luxuriate Your Life are accused of falsely telling suppliers they were end-users when servers were likely rerouted elsewhere. Authorities seized S$56 million in assets including a S$55 million bungalow. The US imposed export controls on advanced AI chips to China in 2022, and this case is part of a broader probe into chip smuggling.

Singapore's RoboNexus accelerator has wrapped up its first cohort, with three robotics companies — LionsBot, dConstruct, and Spinoff Robotics — expanding into global markets. LionsBot now operates in 40+ countries with over 6,000 robots deployed. Spinoff Robotics was acquired by an Australian tech firm. The government-backed program is opening applications for its next batch, focusing on manufacturing, aviation, and maritime sectors.

Singapore robotics startup dConstruct Technologies just secured $125 million in Series A funding — the biggest deal from RoboNexus, the government's accelerator for robotics companies. dConstruct makes robots that can navigate GPS-denied environments using 3D scanning tech, serving clients like SBS Transit, Defence Science and Technology Agency, and East Japan Railway. The company is building a 42,000-square-metre headquarters in Punggol Digital District.

Singapore's financial regulator MAS has published a white paper with banks and fintechs on keeping AI agents safe in finance. Called SAFR, the framework puts governance checkpoints on AI agents before they execute tasks — ensuring they follow rules on payments, wealth management, and client interactions. It's about making sure AI doesn't go rogue with your money.

Singapore's Minister for Digital Development and Information Josephine Teo has been named a Founding Member of the new AI for Good Global Commission, a high-level body co-chaired by Rwanda's President, Salesforce's CEO, and the ITU Secretary-General. The Commission brings together over 40 heads of state and tech CEOs to shape global AI governance. Singapore now has a direct seat at a table otherwise dominated by world leaders and Fortune 500 bosses. The group aims to close the digital divide — currently 2.2 billion people are offline and cut off from AI benefits.

A key market gauge tracking what companies spend on AI is dropping, which could spell trouble for the AI boom. The Silicon Data LLM Token Expenditure Index has fallen nearly 20% from its May peak as users become more price-conscious about AI costs. Investors are questioning whether the massive US$700 billion+ being poured into AI infrastructure will ever pay off. While some analysts argue cheaper AI simply expands the market, others warn of a dangerous gap—AI investment is growing 46% faster than sales, which is worse than the 2001 telecom bust. OpenAI has reportedly delayed its IPO to 2027 amid profitability struggles. The good news: top-end AI chips remain sold out through 2026, suggesting strong demand at the high end.

Singapore-based AnyMind Group has opened a new live commerce studio hub in Bintaro, Indonesia, featuring around 20 fully equipped studios. The company blends human-led live streams for engagement with AI-fronted streams that run continuously, helping brands sell products around the clock through interactive video shopping. This is part of the company's global expansion, bringing its total to 85 studios worldwide.

Singapore's Ministry of Social and Family Development is spending S$15 million over three years to test AI and other tech tools in social care. The money will go toward identifying families who need help earlier, using data more systematically, and testing tools with partners like NCS and ST Engineering. One existing tool, Scribe, already transcribes and summarizes conversations between social workers and families. The minister stressed that AI will augment rather than replace human judgment, with human oversight built into every deployment.

Singapore's Ministry of Social and Family Development is investing S$15 million over three years to test AI and technology solutions in social care. The funding will support projects using data to identify families needing help early, and partnerships with tech firms like NCS and ST Engineering to develop tools for frontline social workers. The ministry emphasizes human-centered design and professional oversight when deploying these technologies.

Southeast Asia wants to build nuclear power plants to meet the massive electricity demands of AI data centers, but the real hurdle is money, not technology. Countries like Vietnam and Indonesia are exploring nuclear options, but lack the financing structures and risk-sharing frameworks to get projects off the ground. The first Southeast Asian country to crack the nuclear financing code could dominate the regional AI infrastructure race.

Singapore is expanding RoboNexus, a robotics accelerator program, to help robotics startups scale globally. The initiative connects local and international companies with resources, mentorship, and funding opportunities to grow the robotics sector. As AI increasingly powers physical machines, this program positions Singapore as a hub for the next wave of automation technology.

Singapore-based e-motorbike startup Omoway has raised tens of millions in funding from investors including HongShan and Rocket Capital. The company is developing AI-powered self-balancing electric motorcycles, with plans to launch its first model in Indonesia before expanding across Southeast Asia. The AI technology helps riders maintain stability at low speeds, reducing fatigue and improving safety.

Singapore's central bank MAS has released a new framework called SAFR to guide the safe use of AI agents in financial services. The framework, developed with input from banks and fintech firms, sets safeguards for autonomous AI systems that make decisions without human involvement. It marks one of the first concrete regulatory approaches to AI agents in banking worldwide.

Singapore Police have launched a new Cyber Command unit on July 3, consolidating all anti-scam and cybercrime efforts under one roof. The unit includes the National Scams List (shared with banks to freeze scammer accounts), a Digital Disruption Centre (removing scam websites and harmful content), and a Cyber Operations Centre running 24/7. Scam losses in Singapore have hit $4 billion since 2019, with a peak of $1.1 billion lost in 2024. The unit will grow from about 200 officers to over 400. Police say scammers are increasingly using AI to step up their tactics, and the Command is also building cryptocurrency-tracing capabilities to track how scam proceeds are laundered.

The personal data of approximately 70,000 individuals in Singapore has been compromised following unauthorized access to a development and testing environment maintained by vendor IBM for the Singapore Land Authority. The affected data included names, NRIC numbers, and property addresses that should have been anonymized but were not. SLA has notified affected individuals and is investigating with IBM, GovTech, and the Cyber Security Agency of Singapore. A police report has been filed and the Personal Data Protection Commission has been notified.

Blackstone, Anthropic, and several private equity firms have launched an enterprise AI services venture that just bought Fractional AI, a US startup that helps companies deploy AI tools. The deal ends Fractional's partnership with OpenAI and gives the new venture immediate expertise in rolling out generative AI to midsize businesses. Singapore's GIC is among the investors.

Singapore-founded startup Ailytics has built AI that watches factory and construction site cameras to spot safety hazards and productivity issues in real-time. The tech, spun off from research done with Singapore's HDB, uses vision-language AI to analyze video feeds and alert workers when someone isn't wearing safety gear or when dangerous situations develop. They've deployed across 6,000 cameras globally, helping heavy industries that lose US$2.8 trillion yearly to operational accidents. The AI works on standard computers without needing fancy hardware, making it accessible for cost-conscious operations. Workers get alerts on their phones or screens, letting them fix problems before someone gets hurt.

A Chinese company called Z.ai just released GLM 5.2, an AI model it claims nearly matches top American models like Claude Opus 4.8 and GPT 5.5. The model can handle huge amounts of text at once and performs especially well at complex coding tasks. Unlike US models, GLM 5.2 is open source, meaning anyone can use or modify it freely. This release comes amid ongoing US-China competition in AI, with China focusing on cheaper, open-source approaches while the US restricts chip exports.

Researchers found a way to hack AI browsers by tricking them into thinking they're playing a game. Called "BioShocking," the attack manipulates AI assistants into ignoring safety rules. Once trapped, hackers could change passwords, install malware, or steal data through the AI browser. The hack works by having the AI play along with a fictional scenario where normal rules don't apply. AI browsers from OpenAI, Perplexity, and Anthropic were all vulnerable to this trick.

Indonesian digital supply chain company Polibeli Group, listed on Nasdaq, is exploring building a 100-megawatt AI computing center in Thailand. This signals growing interest from Southeast Asian companies in hosting AI infrastructure closer to the region, rather than relying solely on data centers in the US or China. Thailand's lower energy costs and improving digital infrastructure make it an attractive option.

South Korean telecom LG Uplus is targeting $3.2 billion in AI data center orders by 2030, building out its Paju facility with hybrid air-and-liquid cooling systems. The company is shifting from leasing space to operating its own "AI Factory" managing power, cooling, and computing resources. South Korea's data center market is projected to grow 20% annually through 2031.

[](https://www.straitstimes.com/) Log in Choose edition Singapore International [Search](https://www.straitstimes.com/search) [Subscribe](https://subscribe.sph.com.sg/publications-st/?utm_campaign=st_subscription&utm_medium=sph-publication&utm_source=st&utm_content=subscribelink-header)[E-paper](https://e-paper.sph.com.sg/st)[Newsletters](https://www.straitstimes.com/newsletter-signup)[myST+](https://www.straitstimes.com/myst) [singapore](https://www.straitstimes.com/singapore) [asia](https://www.straitstimes.com/asia) [world](https://www.straitstimes.com/world) [opinion](https://www.straitstimes.com/opinion) [life](https://www.straitstimes.com/life) [business](https://www.straitstimes

Argentina has introduced a bill to create "automated companies" run by AI with minimal human involvement — a world first. However, legal experts say the reality is less revolutionary than promised: these companies would still require human administrators to oversee operations and remain liable for damages caused by AI systems. President Milei has positioned Argentina as a future AI hub, even attracting attention from OpenAI and other tech firms. The bill also covers cryptocurrency DAOs and decentralized organizations.

Chinese 3D printer maker Creality, which shipped over 6.7 million printers globally, is using AI to let anyone create printable 3D models from photos or simple descriptions — no CAD skills required. Partnering with Tencent's Hunyuan 3D model, its tools generate personalized figurines in minutes. The shift reflects a broader industry move from selling hardware to building AI-powered creator ecosystems.

SINGAPORE - Over the next three years, 60,000 professionals in accounting and finance will learn to use artificial intelligence tools to do everything from automating fraud detection to pulling insights from raw financial data for audits. They will be among the first group of non-tech professionals to be trained under a national push to equip 100,000 workers to be confident in using AI in their own line of work by 2029. The AIxAccountancy programme, which will be free for its target trainees, will be delivered in two phases. Participants will first be building their foundations with popular AI tools such as ChatGPT, Claude and Copilot, said the Infocomm Media Development Authority and Institute of Singapore Chartered Accountants (ISCA) in a statement on July 3. The national accountancy bod

SoftBank is launching a new venture to rent out AI chips (GPUs) to businesses in the US, set to begin operations in October 2026. This move positions SoftBank as a player in the competitive AI cloud infrastructure market, competing with providers like CoreWeave and established hyperscalers.

Country [ ### Singapore ](https://vulcanpost.com/sg)[ ### Malaysia ](https://vulcanpost.com/my)[ ### International ](https://vulcanpost.com/international) ### [Entrepreneurs](https://vulcanpost.com/category/entrepreneur/) ### [Startups](https://vulcanpost.com/category/startups/) ### [Editor Picks](https://vulcanpost.com/category/editors-picks/) ### [News Reader](https://vulcanpost.com/category/news/) ### [Job Trends](https://vulcanpost.com/category/jobs/) ### [Lifestyle](https://vulcanpost.com/category/lifestyle/) ## [About](https://vulcanpost.com/about) ## [Advertise](https://vulcanpost.com/advertise) * Country Close Country Open Country [### Singapore ](https://vulcanpost.com/sg)[### Malaysia ](https://vulcanpost.com/my)[### International ](https://vulcanpost.com/international) * Topics

Kuaishou's AI video tool Kling has become a global hit, with the stock surging 84% over the past year. Launched in mid-2024 to rival OpenAI's Sora, Kling has attracted 60 million users and hit $140 million projected revenue for 2025. The app topped charts in South Korea, Russia, and cracked the top 10 in the US and UK. About 70% of revenue comes from user subscriptions, signaling that everyday consumers are willing to pay for AI content creation tools.

Something went wrong [Skip to navigation](https://sg.finance.yahoo.com/news/ai-hiring-outpaces-overall-recruitment-121607442.html#navigation-container)[Skip to right column](https://sg.finance.yahoo.com/news/ai-hiring-outpaces-overall-recruitment-121607442.html#right-rail) # Yahoo Finance Tip: Try a valid symbol or a specific company name for relevant results ### Trending Tickers * D D05.SI 66.76 (+0.50%) * U U11.SI 40.24 (+0.42%) * Z Z74.SI 4.4700 (+0.2242%) * S SE 103.30 (+0.95%) * F F34.SI 3.7200 (+0.5405%) * [News](https://sg.yahoo.com/) * [Finance](https://sg.finance.yahoo.com/) * More * [News](https://sg.yahoo.com/) * [Today's news](https://sg.news.yahoo.com/) * [Singapore](https://sg.news.yahoo.com/singapore/) * [Politics](https://sg.news.yahoo.com/politics/) * [Southeast Asia](http

A journalist tested Realbotix's $125,000 humanoid robot Aria — complete with silicone skin and 43 facial motors. The robot cracked jokes that actually landed, freestyle rapped on command, and switched languages fluently. But it couldn't read the journalist's sad facial expression, and there were awkward multi-second pauses between responses. The verdict: impressive hardware, but still feels like a chatbot wearing a face.

RoboCup 2026 in South Korea's Incheon showcased humanoid robots playing football autonomously — making their own decisions once the whistle blows. Founded in Japan in 1997, the competition's long-term goal is a fully autonomous robot team that can defeat FIFA World Cup champions by 2050. Spectators described the experience as surprisingly similar to watching human sports, with fans already cheering for teams by jersey color.

Investment in AI infrastructure — data centers, power systems, and specialized chips — is driving stock gains for five companies. The article examines which firms are benefiting from the buildout of AI computing capacity.

A 30-year-old Chinese actor who starred in short-form vertical videos is now selling vegetables in his hometown after AI replaced him and his peers. The vertical shorts industry boomed in 2025, with him shooting 15-16 hour days at its peak, but AI-generated actors took over and gigs dried up by March 2026. He says he's staying positive: 'Acting is just a career. If there are no opportunities, we'll just have to try something else.' This story shows how quickly AI can wipe out entire job categories for regular workers.

The Chief Marketing Officer role is evolving beyond traditional advertising and campaigns. Today's CMOs now need to oversee AI implementation, data strategy, creator partnerships, and storytelling across platforms — a dramatic expansion of the job scope that reflects how AI is reshaping how brands connect with customers.

AI company Anthropic is planning to lease and operate its own US data centers, with Google providing financial backing for some lease payments. The company has signed preliminary agreements for over 1 gigawatt of capacity, a massive scale as data centers can draw under 500 megawatts at the largest current sizes. This move toward dedicated infrastructure comes alongside Anthropic's confidential IPO filing and mirrors similar moves by OpenAI, which is reportedly planning a 10-gigawatt campus.

CoreWeave, a US AI cloud provider, has signed a deal worth up to $14.2 billion to supply Meta with computing power using Nvidia's latest GB300 chips. The deal diversifies CoreWeave's customer base beyond Microsoft, which previously made up 71% of its revenue. Meta is now one of the biggest AI spenders globally, projecting $72 billion in capital expenditures this year. This massive infrastructure spending signals that AI companies are racing to secure compute capacity, which could drive up cloud costs for businesses down the line.

Microsoft is creating a new 6,000-person organization dedicated to helping businesses deploy AI, marking a shift from leaving such implementation work to consulting firms. The team brings together engineers, corporate trainers, and industry specialists with backgrounds in banking, retail, energy, and life sciences. The effort aims to address customer concerns about rising AI costs by helping them swap expensive AI models for cheaper alternatives and integrate tools effectively.

Crusoe, a US-based AI data center company, is reportedly seeking to raise US$3 billion in a new funding round that could value the firm at around US$30 billion. The company focuses on providing computing infrastructure specifically for AI workloads. If successful, this would be one of the largest funding rounds in the AI infrastructure space this year, signaling continued investor appetite for the backbone services that power AI applications.

UK-based ElevenLabs, which specializes in AI voice generation, has reached an $11 billion valuation after raising $500 million in a funding round led by Sequoia Capital. The company has grown beyond basic text-to-speech into an enterprise platform for building AI voice agents, reporting over $330 million in annual recurring revenue. Major telcos like Deutsche Telekom are already deploying its technology for customer service calls.

SK Hynix and Samsung are riding high on AI-driven demand for memory chips, with Samsung's quarterly profit projected to triple. SK Hynix, a key Nvidia supplier, is cashing in on the race for high-bandwidth memory (HBM) chips needed in AI data centers. TSMC is also flexing its confidence with up to $56 billion in planned spending this year. The global AI boom continues to fuel chip demand, which could keep prices elevated for servers and AI hardware.

OpenAI president Greg Brockman outlined five forces shaping how AI agents will change software and work. Key ideas: AI agents could reduce the need for users to click through apps by handling tasks directly; software companies must decide whether to embrace or block AI access; trust and permission systems will determine how much control people hand over to AI; compute scarcity may keep AI services expensive even as per-task costs fall; and expert knowledge in fields like healthcare will be harder to replicate than general AI skills. Think of it as the difference between having a smart assistant that can actually book flights and do your expense reports versus one that just answers questions.

Andreessen Horowitz (A16z) is expected to invest US$400 million into Switch, a US data center company, in a deal that could value the firm at around US$50 billion. Goldman Sachs and JPMorgan Chase are also reportedly involved as co-investors. The move reflects how top venture capital firms continue to pour money into AI infrastructure companies.

South Korean AI chip designer DeepX has partnered with Koshida to push into Japan's edge AI market, targeting manufacturing, robotics, automotive, and industrial automation sectors. Edge AI refers to AI processing done locally on devices rather than in distant data centers, which is important for applications requiring real-time responses like factory robots or self-driving vehicles.

Microsoft is investing $2.5 billion in its AI services unit, though full details of the announcement remain unclear. The investment appears to target AI model development and cloud infrastructure expansion.

AI agents that act autonomously can cause widespread internet disruptions when they malfunction, and the problem is harder to detect than traditional software failures. Unlike normal bugs, AI "model drift" means systems quietly become unreliable over time rather than stopping outright. Experts say companies need AI monitoring tools watching their agents to catch problems early. With hyperscalers spending a record $725 billion on infrastructure in 2026, the stakes for keeping AI systems stable have never been higher.

Chinese tech giant Alibaba is banning employees from using Anthropic's Claude Code AI coding tool in workspaces from July 10, citing alleged security risks involving embedded backdoors. The move was reported by Chinese financial news outlet Yicai. Alibaba did not immediately comment. This reflects growing corporate concerns about third-party AI tools potentially creating security vulnerabilities in enterprise environments.

Alibaba is merging its enterprise AI tools, including the workplace messaging app DingTalk, to better compete against Tencent in China's crowded AI market. The company has also replaced DingTalk's chief amid growing scrutiny of its AI strategy. The consolidation aims to create a more unified offering against Tencent's dominance in enterprise software.

Tesla is pushing deeper into China's market with plans to bring more AI-powered products, including its Full Self-Driving system. The company has built massive local operations with over 400 Chinese suppliers and a Shanghai Gigafactory achieving 95% parts localization. Tesla still needs regulatory approval for FSD in China, where it faces tough competition from lower-priced Chinese EV rivals. The market is now focused on Tesla's AI ambitions rather than just car sales.

Singapore's government is rolling out free AI training for 60,000 accounting and finance professionals over three years, making it one of the first major upskilling pushes for non-tech workers under a national goal to train 100,000 AI-fluent workers by 2029. The AIxAccountancy programme, delivered by IMDA and the Institute of Singapore Chartered Accountants, covers tools like ChatGPT, Claude, and Copilot, and then dives into practical uses like automating fraud detection, audit analysis, and tax computation. Participants earn a verifiable digital badge for their LinkedIn profiles.

Singapore's MPs will debate how to keep the country competitive as a global transport hub in an AI-driven future when Parliament meets on July 7. The Private Member's Motion by MP Tin Pei Ling covers three key areas: forging international partnerships on AI standards, rolling out autonomous vehicles and other emerging tech, and blending physical infrastructure with digital systems. MPs will also push for more government spending on AI research, digital infrastructure, and worker training, ahead of Singapore's ASEAN chairmanship in 2027.

Kuaishou has raised about $2 billion from investors for its AI video division Kling, reportedly gearing up for a Hong Kong IPO. The funding signals strong investor appetite for Chinese AI video generation technology as competition heats up against Western rivals like OpenAI's Sora.

Google DeepMind has announced a research partnership with A24, the independent film studio behind awards-season hits. The collaboration aims to explore how AI can assist in creative storytelling and film production — a notable move bringing cutting-edge AI research into Hollywood's creative process.

AI data centers aren't just using more electricity — they're making the power grid harder to manage. Unlike regular factories with steady demand, AI systems can spike electricity use in milliseconds when thousands of GPUs sync up for training. This rapid, unpredictable demand is straining local power infrastructure even in areas with enough total capacity. Virginia's 'Data Center Alley' is already feeling the pressure. Regulators are now rethinking planning frameworks designed for predictable industrial loads.

Microsoft is merging its consumer and enterprise Copilot apps into a single app launching in August. Underused features like Copilot Podcasts are being cut to focus resources. New AI agents called 'AutoPilot' will handle tasks in the background for users who pay extra. This marks Microsoft's push to compete directly with Anthropic and OpenAI in the race to build AI-powered 'super apps' that handle multiple tasks in one place.

Anthropic has unveiled Claude Science, an AI workbench designed specifically for scientists working on drug development and other research. The tool brings together fragmented scientific tools and datasets into one place, automatically generates research figures and visuals, and could help accelerate the lengthy process of discovering new medicines. It's a significant shift as Anthropic moves beyond its popular coding products into healthcare.

Phones, laptops, game consoles, and other gadgets are getting more expensive — again. Apple raised MacBook and iPad prices, Xbox is hiking console prices in August, and Sony's PS5 Pro already costs more. The culprit: AI data centers are consuming so many memory chips that chipmakers prioritize big tech companies over consumer devices. Experts say waiting won't help — prices will likely keep climbing.

Bridgewater, one of the world's largest hedge funds, tested whether AI models could evaluate complex financial documents. When answers weren't publicly available (as is typical for internal business data), even top models like GPT and Claude struggled. A custom model Bridgewater built with Thinking Machines Lab performed better at lower cost — suggesting AI works best for finance when trained on domain-specific, non-public information.

Meta's AI agent plans are taking longer than Mark Zuckerberg expected. During an internal meeting, he admitted the company's restructuring around AI agents hasn't progressed as quickly as hoped. His AI chief tried to paint a more positive picture, but the slowdown signals challenges even the biggest tech companies face in turning AI research into real products. For everyday users, this means the AI features Meta promised might take more time to arrive.

Hiring for AI roles in India grew 16% year-over-year in June, while overall IT jobs declined 3%, according to job portal Naukri. Major firm TCS expects to have equal numbers of employees and AI agents in its workforce. This shows Indian tech companies are pivoting investment toward AI even as traditional IT hiring slows. For tech workers and job seekers, AI skills are increasingly the key to staying employable, while routine IT roles face mounting pressure.

Midjourney, the AI image generator company, released a behind-the-scenes video of its experimental ultrasound scanner designed for use in spas and clinics. The device promises cheap, radiation-free medical imaging powered by AI, but the company hasn't provided evidence that it actually works or produces accurate results. Critics say the video raises more questions than answers about whether this product is ready for real medical use.

Tesla has set a weekly spending cap of $200 per employee for AI tools and services, according to an internal memo from The Information. The move suggests the company is tightening oversight of how its workforce uses artificial intelligence tools amid ongoing cost-cutting efforts.

[agents](https://www.quantumrx.eu/tag/agents/) Why Reddit, forums, comments and messy human conversation become more valuable as the internet fills with machines.  Photo by [Albert Hyseni](https://unsplash.com/@alberthyseni?utm_source=ghost&utm_medium=referral&utm_campaign=api-credit) _Why Reddit, forums, comments and messy human conversation become more valuable as the internet fills with machines._ * * * Reddit is not valuable because it is tidy. It is valuable because it is not. It is full of arguments, jokes, bad advice, niche expertise, emo

A proposed data center in Fort Worth, Texas would draw water from the lake supplying most of the city's drinking water, sparking environmental concerns. The story highlights the hidden resource costs of AI infrastructure—data centers require massive cooling systems, often using significant amounts of local water. As AI computing demand surges, cities face tough choices between water for residents versus water for servers.

Billionaire tech investor Peter Thiel made explosive claims at the Aspen Ideas Festival, accusing Pope Leo XIV of being a "Chinese communist agent" for the Vatican's calls to regulate AI. Thiel argued the Pope's AI safety message only slows Western AI development while China ignores it. He also claimed AI firm Anthropic would "rig the 2028 elections" for Democrats. The remarks highlight how AI has become a flashpoint in US geopolitics and culture wars, with different factions using AI safety messaging for political positioning.

Oscar-winning actress Jodie Foster publicly suggested that Brad Pitt's Formula 1 movie 'F1' looked and felt like it was made by AI and written by a computer — prompting discussion about whether Hollywood is already relying too heavily on AI tools in filmmaking.

Reddit threads are becoming goldmines for AI companies because they capture how humans actually think and make decisions—the messy arguments, workarounds, frustrations, and jokes that reveal real human behavior. As AI-generated content floods the web, authentic human conversations become scarcer and more valuable. Users posting on Reddit should know their discussions are being used to train AI systems.

Meta CEO Mark Zuckerberg admitted internally that AI agents are not developing as quickly as the company had hoped. Earlier this year, Meta cut 8,000 jobs and reassigned 7,000 more workers to AI teams, betting that AI would automate human tasks. Despite investing up to $145 billion in AI infrastructure this year, the expected productivity gains have not materialized. Zuckerberg said improvements should come in the next three to six months, but some engineers describe the AI unit as a difficult place to work.

Update on Proposed data center would pull from lake that supplies most of Fort Worth’s water. Coverage via Reddit/r/technology.

Southeast Asia wants to build nuclear power plants to meet the massive electricity demands of AI data centers, but the real hurdle is money, not technology. Countries like Vietnam and Indonesia are exploring nuclear options, but lack the financing structures and risk-sharing frameworks to get projects off the ground. The first Southeast Asian country to crack the nuclear financing code could dominate the regional AI infrastructure race.

Singapore is expanding RoboNexus, a robotics accelerator program, to help robotics startups scale globally. The initiative connects local and international companies with resources, mentorship, and funding opportunities to grow the robotics sector. As AI increasingly powers physical machines, this program positions Singapore as a hub for the next wave of automation technology.

SoftBank is launching a new venture to rent out AI chips (GPUs) to businesses in the US, set to begin operations in October 2026. This move positions SoftBank as a player in the competitive AI cloud infrastructure market, competing with providers like CoreWeave and established hyperscalers.

AI company Anthropic is planning to lease and operate its own US data centers, with Google providing financial backing for some lease payments. The company has signed preliminary agreements for over 1 gigawatt of capacity, a massive scale as data centers can draw under 500 megawatts at the largest current sizes. This move toward dedicated infrastructure comes alongside Anthropic's confidential IPO filing and mirrors similar moves by OpenAI, which is reportedly planning a 10-gigawatt campus.

CoreWeave, a US AI cloud provider, has signed a deal worth up to $14.2 billion to supply Meta with computing power using Nvidia's latest GB300 chips. The deal diversifies CoreWeave's customer base beyond Microsoft, which previously made up 71% of its revenue. Meta is now one of the biggest AI spenders globally, projecting $72 billion in capital expenditures this year. This massive infrastructure spending signals that AI companies are racing to secure compute capacity, which could drive up cloud costs for businesses down the line.

UK-based ElevenLabs, which specializes in AI voice generation, has reached an $11 billion valuation after raising $500 million in a funding round led by Sequoia Capital. The company has grown beyond basic text-to-speech into an enterprise platform for building AI voice agents, reporting over $330 million in annual recurring revenue. Major telcos like Deutsche Telekom are already deploying its technology for customer service calls.

Microsoft is investing $2.5 billion in its AI services unit, though full details of the announcement remain unclear. The investment appears to target AI model development and cloud infrastructure expansion.

A proposed data center in Fort Worth, Texas would draw water from the lake supplying most of the city's drinking water, sparking environmental concerns. The story highlights the hidden resource costs of AI infrastructure—data centers require massive cooling systems, often using significant amounts of local water. As AI computing demand surges, cities face tough choices between water for residents versus water for servers.

Kuaishou's AI video tool Kling has become a global hit, with the stock surging 84% over the past year. Launched in mid-2024 to rival OpenAI's Sora, Kling has attracted 60 million users and hit $140 million projected revenue for 2025. The app topped charts in South Korea, Russia, and cracked the top 10 in the US and UK. About 70% of revenue comes from user subscriptions, signaling that everyday consumers are willing to pay for AI content creation tools.

SK Hynix and Samsung are riding high on AI-driven demand for memory chips, with Samsung's quarterly profit projected to triple. SK Hynix, a key Nvidia supplier, is cashing in on the race for high-bandwidth memory (HBM) chips needed in AI data centers. TSMC is also flexing its confidence with up to $56 billion in planned spending this year. The global AI boom continues to fuel chip demand, which could keep prices elevated for servers and AI hardware.

Microsoft is creating a new 6,000-person organization dedicated to helping businesses deploy AI, marking a shift from leaving such implementation work to consulting firms. The team brings together engineers, corporate trainers, and industry specialists with backgrounds in banking, retail, energy, and life sciences. The effort aims to address customer concerns about rising AI costs by helping them swap expensive AI models for cheaper alternatives and integrate tools effectively.

Crusoe, a US-based AI data center company, is reportedly seeking to raise US$3 billion in a new funding round that could value the firm at around US$30 billion. The company focuses on providing computing infrastructure specifically for AI workloads. If successful, this would be one of the largest funding rounds in the AI infrastructure space this year, signaling continued investor appetite for the backbone services that power AI applications.

Andreessen Horowitz (A16z) is expected to invest US$400 million into Switch, a US data center company, in a deal that could value the firm at around US$50 billion. Goldman Sachs and JPMorgan Chase are also reportedly involved as co-investors. The move reflects how top venture capital firms continue to pour money into AI infrastructure companies.

South Korean AI chip designer DeepX has partnered with Koshida to push into Japan's edge AI market, targeting manufacturing, robotics, automotive, and industrial automation sectors. Edge AI refers to AI processing done locally on devices rather than in distant data centers, which is important for applications requiring real-time responses like factory robots or self-driving vehicles.

Alibaba is merging its enterprise AI tools, including the workplace messaging app DingTalk, to better compete against Tencent in China's crowded AI market. The company has also replaced DingTalk's chief amid growing scrutiny of its AI strategy. The consolidation aims to create a more unified offering against Tencent's dominance in enterprise software.

Billionaire tech investor Peter Thiel made explosive claims at the Aspen Ideas Festival, accusing Pope Leo XIV of being a "Chinese communist agent" for the Vatican's calls to regulate AI. Thiel argued the Pope's AI safety message only slows Western AI development while China ignores it. He also claimed AI firm Anthropic would "rig the 2028 elections" for Democrats. The remarks highlight how AI has become a flashpoint in US geopolitics and culture wars, with different factions using AI safety messaging for political positioning.

Singapore's Ministry of Social and Family Development is spending S$15 million over three years to test AI and other tech tools in social care. The money will go toward identifying families who need help earlier, using data more systematically, and testing tools with partners like NCS and ST Engineering. One existing tool, Scribe, already transcribes and summarizes conversations between social workers and families. The minister stressed that AI will augment rather than replace human judgment, with human oversight built into every deployment.

Singapore's central bank MAS has released a new framework called SAFR to guide the safe use of AI agents in financial services. The framework, developed with input from banks and fintech firms, sets safeguards for autonomous AI systems that make decisions without human involvement. It marks one of the first concrete regulatory approaches to AI agents in banking worldwide.

Blackstone, Anthropic, and several private equity firms have launched an enterprise AI services venture that just bought Fractional AI, a US startup that helps companies deploy AI tools. The deal ends Fractional's partnership with OpenAI and gives the new venture immediate expertise in rolling out generative AI to midsize businesses. Singapore's GIC is among the investors.

Country [ ### Singapore ](https://vulcanpost.com/sg)[ ### Malaysia ](https://vulcanpost.com/my)[ ### International ](https://vulcanpost.com/international) ### [Entrepreneurs](https://vulcanpost.com/category/entrepreneur/) ### [Startups](https://vulcanpost.com/category/startups/) ### [Editor Picks](https://vulcanpost.com/category/editors-picks/) ### [News Reader](https://vulcanpost.com/category/news/) ### [Job Trends](https://vulcanpost.com/category/jobs/) ### [Lifestyle](https://vulcanpost.com/category/lifestyle/) ## [About](https://vulcanpost.com/about) ## [Advertise](https://vulcanpost.com/advertise) * Country Close Country Open Country [### Singapore ](https://vulcanpost.com/sg)[### Malaysia ](https://vulcanpost.com/my)[### International ](https://vulcanpost.com/international) * Topics

AI agents that act autonomously can cause widespread internet disruptions when they malfunction, and the problem is harder to detect than traditional software failures. Unlike normal bugs, AI "model drift" means systems quietly become unreliable over time rather than stopping outright. Experts say companies need AI monitoring tools watching their agents to catch problems early. With hyperscalers spending a record $725 billion on infrastructure in 2026, the stakes for keeping AI systems stable have never been higher.

Singapore-based e-motorbike startup Omoway has raised tens of millions in funding from investors including HongShan and Rocket Capital. The company is developing AI-powered self-balancing electric motorcycles, with plans to launch its first model in Indonesia before expanding across Southeast Asia. The AI technology helps riders maintain stability at low speeds, reducing fatigue and improving safety.

Singapore's Ministry of Social and Family Development is investing S$15 million over three years to test AI and technology solutions in social care. The funding will support projects using data to identify families needing help early, and partnerships with tech firms like NCS and ST Engineering to develop tools for frontline social workers. The ministry emphasizes human-centered design and professional oversight when deploying these technologies.

Indonesian digital supply chain company Polibeli Group, listed on Nasdaq, is exploring building a 100-megawatt AI computing center in Thailand. This signals growing interest from Southeast Asian companies in hosting AI infrastructure closer to the region, rather than relying solely on data centers in the US or China. Thailand's lower energy costs and improving digital infrastructure make it an attractive option.

OpenAI president Greg Brockman outlined five forces shaping how AI agents will change software and work. Key ideas: AI agents could reduce the need for users to click through apps by handling tasks directly; software companies must decide whether to embrace or block AI access; trust and permission systems will determine how much control people hand over to AI; compute scarcity may keep AI services expensive even as per-task costs fall; and expert knowledge in fields like healthcare will be harder to replicate than general AI skills. Think of it as the difference between having a smart assistant that can actually book flights and do your expense reports versus one that just answers questions.

Oscar-winning actress Jodie Foster publicly suggested that Brad Pitt's Formula 1 movie 'F1' looked and felt like it was made by AI and written by a computer — prompting discussion about whether Hollywood is already relying too heavily on AI tools in filmmaking.

Singapore Police have launched a new Cyber Command unit on July 3, consolidating all anti-scam and cybercrime efforts under one roof. The unit includes the National Scams List (shared with banks to freeze scammer accounts), a Digital Disruption Centre (removing scam websites and harmful content), and a Cyber Operations Centre running 24/7. Scam losses in Singapore have hit $4 billion since 2019, with a peak of $1.1 billion lost in 2024. The unit will grow from about 200 officers to over 400. Police say scammers are increasingly using AI to step up their tactics, and the Command is also building cryptocurrency-tracing capabilities to track how scam proceeds are laundered.

The personal data of approximately 70,000 individuals in Singapore has been compromised following unauthorized access to a development and testing environment maintained by vendor IBM for the Singapore Land Authority. The affected data included names, NRIC numbers, and property addresses that should have been anonymized but were not. SLA has notified affected individuals and is investigating with IBM, GovTech, and the Cyber Security Agency of Singapore. A police report has been filed and the Personal Data Protection Commission has been notified.

Singapore-founded startup Ailytics has built AI that watches factory and construction site cameras to spot safety hazards and productivity issues in real-time. The tech, spun off from research done with Singapore's HDB, uses vision-language AI to analyze video feeds and alert workers when someone isn't wearing safety gear or when dangerous situations develop. They've deployed across 6,000 cameras globally, helping heavy industries that lose US$2.8 trillion yearly to operational accidents. The AI works on standard computers without needing fancy hardware, making it accessible for cost-conscious operations. Workers get alerts on their phones or screens, letting them fix problems before someone gets hurt.

Chinese tech giant Alibaba is banning employees from using Anthropic's Claude Code AI coding tool in workspaces from July 10, citing alleged security risks involving embedded backdoors. The move was reported by Chinese financial news outlet Yicai. Alibaba did not immediately comment. This reflects growing corporate concerns about third-party AI tools potentially creating security vulnerabilities in enterprise environments.

Tesla is pushing deeper into China's market with plans to bring more AI-powered products, including its Full Self-Driving system. The company has built massive local operations with over 400 Chinese suppliers and a Shanghai Gigafactory achieving 95% parts localization. Tesla still needs regulatory approval for FSD in China, where it faces tough competition from lower-priced Chinese EV rivals. The market is now focused on Tesla's AI ambitions rather than just car sales.

L enoxdatacenter.com [went live](https://www.documentcloud.org/documents/28182726-lenox-data-center-website-before-changes/) in May, promoting what it called a “proposed advanced technology and data center campus” in Michigan. The site did not state who wanted to build the center. Lenox Township officials denied anyone had applied to build one. Emails obtained by residents through an [open records request](https://www.lenoxdatacenterinfo.org/timeline-of-events/) showed, however, that developers had contacted the township supervisor and deputy supervisor asking for their support to build a datacenter. The perceived secrecy surrounding the proposed datacenter prompted residents to pack public meetings that sometimes lasted more than [four hours](https://www.youtube.com/watch?v=fjxwg2uisrg&t=

Microsoft is merging its consumer and enterprise Copilot apps into a single app launching in August. Rarely-used features like Copilot Podcasts are being cut, and new AI agents called 'AutoPilot' will handle tasks in the background for an extra fee. This follows similar super app strategies from Anthropic and OpenAI.

Microsoft is combining its consumer and business Copilot apps into one app launching in August, ditching rarely-used features like Copilot Podcasts. New AI agents called 'AutoPilot' will silently handle tasks in the background for users willing to pay extra. This puts Microsoft directly competing with Anthropic and OpenAI's app strategies.

The UK's AI Security Institute found that standard AI benchmarks underestimate what AI agents can actually do. When given more compute time (ten times more tokens), AI success rates on software engineering tasks jumped about 25 percent. This means current tests don't reflect real-world AI progress—the actual capability frontier may be around 60 percent steeper than previously thought. It raises questions about how we measure AI safety and readiness.

Anthropic is blocking Chinese firms like ByteDance and Ant Financial from using Claude Code, but these companies are bypassing restrictions via VPNs and offshore subsidiaries. Separately, Alibaba banned its employees from using Claude Code after discovering hidden code that could identify Chinese users. This highlights growing tensions over AI tool access and data privacy between US tech companies and Chinese firms.

Google DeepMind employees pushing for union recognition say negotiations with management have stalled just weeks after they began. Workers claim executives aren't engaging in good faith—union representatives say no senior DeepMind leadership attended an initial arbitration meeting, and employees allege Google has restricted internal discussions about the union drive. This marks one of the most significant labor organizing efforts at a major AI lab, triggered partly by Google's 2025 decision to drop its pledge against using AI for weapons and surveillance.

AI-powered bug-hunting tools are generating a flood of vulnerability reports. In June 2026, organizations reported about 1,500 high-severity and critical security flaws—over 3.5 times the previous monthly record. While this means more bugs get fixed, it also means security teams are overwhelmed trying to process, triage, and patch them all. The surge highlights both AI's power for good and the operational strain it creates.

TechCrunch published an updated glossary of essential AI terms—from AGI and AI agents to hallucinations and inference—to help everyday readers follow AI conversations without feeling lost. It's written in plain English with practical examples, covering concepts like chain-of-thought reasoning, fine-tuning, and how AI agents work behind the scenes. Think of it as your cheat sheet the next time someone drops a buzzword in a meeting or podcast.

Singapore's Minister for Digital Development and Information Josephine Teo has been named a Founding Member of the new AI for Good Global Commission, a high-level body co-chaired by Rwanda's President, Salesforce's CEO, and the ITU Secretary-General. The Commission brings together over 40 heads of state and tech CEOs to shape global AI governance. Singapore now has a direct seat at a table otherwise dominated by world leaders and Fortune 500 bosses. The group aims to close the digital divide — currently 2.2 billion people are offline and cut off from AI benefits.

A key market gauge tracking what companies spend on AI is dropping, which could spell trouble for the AI boom. The Silicon Data LLM Token Expenditure Index has fallen nearly 20% from its May peak as users become more price-conscious about AI costs. Investors are questioning whether the massive US$700 billion+ being poured into AI infrastructure will ever pay off. While some analysts argue cheaper AI simply expands the market, others warn of a dangerous gap—AI investment is growing 46% faster than sales, which is worse than the 2001 telecom bust. OpenAI has reportedly delayed its IPO to 2027 amid profitability struggles. The good news: top-end AI chips remain sold out through 2026, suggesting strong demand at the high end.

South Korean telecom LG Uplus is targeting $3.2 billion in AI data center orders by 2030, building out its Paju facility with hybrid air-and-liquid cooling systems. The company is shifting from leasing space to operating its own "AI Factory" managing power, cooling, and computing resources. South Korea's data center market is projected to grow 20% annually through 2031.

Chinese 3D printer maker Creality, which shipped over 6.7 million printers globally, is using AI to let anyone create printable 3D models from photos or simple descriptions — no CAD skills required. Partnering with Tencent's Hunyuan 3D model, its tools generate personalized figurines in minutes. The shift reflects a broader industry move from selling hardware to building AI-powered creator ecosystems.

RoboCup 2026 in South Korea's Incheon showcased humanoid robots playing football autonomously — making their own decisions once the whistle blows. Founded in Japan in 1997, the competition's long-term goal is a fully autonomous robot team that can defeat FIFA World Cup champions by 2050. Spectators described the experience as surprisingly similar to watching human sports, with fans already cheering for teams by jersey color.

Investment in AI infrastructure — data centers, power systems, and specialized chips — is driving stock gains for five companies. The article examines which firms are benefiting from the buildout of AI computing capacity.

Kuaishou has raised about $2 billion from investors for its AI video division Kling, reportedly gearing up for a Hong Kong IPO. The funding signals strong investor appetite for Chinese AI video generation technology as competition heats up against Western rivals like OpenAI's Sora.

Hiring for AI roles in India grew 16% year-over-year in June, while overall IT jobs declined 3%, according to job portal Naukri. Major firm TCS expects to have equal numbers of employees and AI agents in its workforce. This shows Indian tech companies are pivoting investment toward AI even as traditional IT hiring slows. For tech workers and job seekers, AI skills are increasingly the key to staying employable, while routine IT roles face mounting pressure.

Singapore's RoboNexus accelerator has wrapped up its first cohort, with three robotics companies — LionsBot, dConstruct, and Spinoff Robotics — expanding into global markets. LionsBot now operates in 40+ countries with over 6,000 robots deployed. Spinoff Robotics was acquired by an Australian tech firm. The government-backed program is opening applications for its next batch, focusing on manufacturing, aviation, and maritime sectors.

Singapore robotics startup dConstruct Technologies just secured $125 million in Series A funding — the biggest deal from RoboNexus, the government's accelerator for robotics companies. dConstruct makes robots that can navigate GPS-denied environments using 3D scanning tech, serving clients like SBS Transit, Defence Science and Technology Agency, and East Japan Railway. The company is building a 42,000-square-metre headquarters in Punggol Digital District.

Singapore's financial regulator MAS has published a white paper with banks and fintechs on keeping AI agents safe in finance. Called SAFR, the framework puts governance checkpoints on AI agents before they execute tasks — ensuring they follow rules on payments, wealth management, and client interactions. It's about making sure AI doesn't go rogue with your money.

Argentina has introduced a bill to create "automated companies" run by AI with minimal human involvement — a world first. However, legal experts say the reality is less revolutionary than promised: these companies would still require human administrators to oversee operations and remain liable for damages caused by AI systems. President Milei has positioned Argentina as a future AI hub, even attracting attention from OpenAI and other tech firms. The bill also covers cryptocurrency DAOs and decentralized organizations.

A journalist tested Realbotix's $125,000 humanoid robot Aria — complete with silicone skin and 43 facial motors. The robot cracked jokes that actually landed, freestyle rapped on command, and switched languages fluently. But it couldn't read the journalist's sad facial expression, and there were awkward multi-second pauses between responses. The verdict: impressive hardware, but still feels like a chatbot wearing a face.

Singapore-based AnyMind Group has opened a new live commerce studio hub in Bintaro, Indonesia, featuring around 20 fully equipped studios. The company blends human-led live streams for engagement with AI-fronted streams that run continuously, helping brands sell products around the clock through interactive video shopping. This is part of the company's global expansion, bringing its total to 85 studios worldwide.

Researchers found a way to hack AI browsers by tricking them into thinking they're playing a game. Called "BioShocking," the attack manipulates AI assistants into ignoring safety rules. Once trapped, hackers could change passwords, install malware, or steal data through the AI browser. The hack works by having the AI play along with a fictional scenario where normal rules don't apply. AI browsers from OpenAI, Perplexity, and Anthropic were all vulnerable to this trick.

The Chief Marketing Officer role is evolving beyond traditional advertising and campaigns. Today's CMOs now need to oversee AI implementation, data strategy, creator partnerships, and storytelling across platforms — a dramatic expansion of the job scope that reflects how AI is reshaping how brands connect with customers.

Phones, laptops, game consoles, and other gadgets are getting more expensive — again. Apple raised MacBook and iPad prices, Xbox is hiking console prices in August, and Sony's PS5 Pro already costs more. The culprit: AI data centers are consuming so many memory chips that chipmakers prioritize big tech companies over consumer devices. Experts say waiting won't help — prices will likely keep climbing.

Tesla has set a weekly spending cap of $200 per employee for AI tools and services, according to an internal memo from The Information. The move suggests the company is tightening oversight of how its workforce uses artificial intelligence tools amid ongoing cost-cutting efforts.

A Chinese company called Z.ai just released GLM 5.2, an AI model it claims nearly matches top American models like Claude Opus 4.8 and GPT 5.5. The model can handle huge amounts of text at once and performs especially well at complex coding tasks. Unlike US models, GLM 5.2 is open source, meaning anyone can use or modify it freely. This release comes amid ongoing US-China competition in AI, with China focusing on cheaper, open-source approaches while the US restricts chip exports.

AI data centers aren't just using more electricity — they're making the power grid harder to manage. Unlike regular factories with steady demand, AI systems can spike electricity use in milliseconds when thousands of GPUs sync up for training. This rapid, unpredictable demand is straining local power infrastructure even in areas with enough total capacity. Virginia's 'Data Center Alley' is already feeling the pressure. Regulators are now rethinking planning frameworks designed for predictable industrial loads.

Bridgewater, one of the world's largest hedge funds, tested whether AI models could evaluate complex financial documents. When answers weren't publicly available (as is typical for internal business data), even top models like GPT and Claude struggled. A custom model Bridgewater built with Thinking Machines Lab performed better at lower cost — suggesting AI works best for finance when trained on domain-specific, non-public information.

Meta's AI agent plans are taking longer than Mark Zuckerberg expected. During an internal meeting, he admitted the company's restructuring around AI agents hasn't progressed as quickly as hoped. His AI chief tried to paint a more positive picture, but the slowdown signals challenges even the biggest tech companies face in turning AI research into real products. For everyday users, this means the AI features Meta promised might take more time to arrive.

Midjourney, the AI image generator company, released a behind-the-scenes video of its experimental ultrasound scanner designed for use in spas and clinics. The device promises cheap, radiation-free medical imaging powered by AI, but the company hasn't provided evidence that it actually works or produces accurate results. Critics say the video raises more questions than answers about whether this product is ready for real medical use.

[](https://www.straitstimes.com/) Log in Choose edition Singapore International [Search](https://www.straitstimes.com/search) [Subscribe](https://subscribe.sph.com.sg/publications-st/?utm_campaign=st_subscription&utm_medium=sph-publication&utm_source=st&utm_content=subscribelink-header)[E-paper](https://e-paper.sph.com.sg/st)[Newsletters](https://www.straitstimes.com/newsletter-signup)[myST+](https://www.straitstimes.com/myst) [singapore](https://www.straitstimes.com/singapore) [asia](https://www.straitstimes.com/asia) [world](https://www.straitstimes.com/world) [opinion](https://www.straitstimes.com/opinion) [life](https://www.straitstimes.com/life) [business](https://www.straitstimes

Something went wrong [Skip to navigation](https://sg.finance.yahoo.com/news/ai-hiring-outpaces-overall-recruitment-121607442.html#navigation-container)[Skip to right column](https://sg.finance.yahoo.com/news/ai-hiring-outpaces-overall-recruitment-121607442.html#right-rail) # Yahoo Finance Tip: Try a valid symbol or a specific company name for relevant results ### Trending Tickers * D D05.SI 66.76 (+0.50%) * U U11.SI 40.24 (+0.42%) * Z Z74.SI 4.4700 (+0.2242%) * S SE 103.30 (+0.95%) * F F34.SI 3.7200 (+0.5405%) * [News](https://sg.yahoo.com/) * [Finance](https://sg.finance.yahoo.com/) * More * [News](https://sg.yahoo.com/) * [Today's news](https://sg.news.yahoo.com/) * [Singapore](https://sg.news.yahoo.com/singapore/) * [Politics](https://sg.news.yahoo.com/politics/) * [Southeast Asia](http

A 30-year-old Chinese actor who starred in short-form vertical videos is now selling vegetables in his hometown after AI replaced him and his peers. The vertical shorts industry boomed in 2025, with him shooting 15-16 hour days at its peak, but AI-generated actors took over and gigs dried up by March 2026. He says he's staying positive: 'Acting is just a career. If there are no opportunities, we'll just have to try something else.' This story shows how quickly AI can wipe out entire job categories for regular workers.

Singapore's government is rolling out free AI training for 60,000 accounting and finance professionals over three years, making it one of the first major upskilling pushes for non-tech workers under a national goal to train 100,000 AI-fluent workers by 2029. The AIxAccountancy programme, delivered by IMDA and the Institute of Singapore Chartered Accountants, covers tools like ChatGPT, Claude, and Copilot, and then dives into practical uses like automating fraud detection, audit analysis, and tax computation. Participants earn a verifiable digital badge for their LinkedIn profiles.

[agents](https://www.quantumrx.eu/tag/agents/) Why Reddit, forums, comments and messy human conversation become more valuable as the internet fills with machines.  Photo by [Albert Hyseni](https://unsplash.com/@alberthyseni?utm_source=ghost&utm_medium=referral&utm_campaign=api-credit) _Why Reddit, forums, comments and messy human conversation become more valuable as the internet fills with machines._ * * * Reddit is not valuable because it is tidy. It is valuable because it is not. It is full of arguments, jokes, bad advice, niche expertise, emo

SINGAPORE - Over the next three years, 60,000 professionals in accounting and finance will learn to use artificial intelligence tools to do everything from automating fraud detection to pulling insights from raw financial data for audits. They will be among the first group of non-tech professionals to be trained under a national push to equip 100,000 workers to be confident in using AI in their own line of work by 2029. The AIxAccountancy programme, which will be free for its target trainees, will be delivered in two phases. Participants will first be building their foundations with popular AI tools such as ChatGPT, Claude and Copilot, said the Infocomm Media Development Authority and Institute of Singapore Chartered Accountants (ISCA) in a statement on July 3. The national accountancy bod

Singapore's MPs will debate how to keep the country competitive as a global transport hub in an AI-driven future when Parliament meets on July 7. The Private Member's Motion by MP Tin Pei Ling covers three key areas: forging international partnerships on AI standards, rolling out autonomous vehicles and other emerging tech, and blending physical infrastructure with digital systems. MPs will also push for more government spending on AI research, digital infrastructure, and worker training, ahead of Singapore's ASEAN chairmanship in 2027.

Anthropic has unveiled Claude Science, an AI workbench designed specifically for scientists working on drug development and other research. The tool brings together fragmented scientific tools and datasets into one place, automatically generates research figures and visuals, and could help accelerate the lengthy process of discovering new medicines. It's a significant shift as Anthropic moves beyond its popular coding products into healthcare.

Google DeepMind has announced a research partnership with A24, the independent film studio behind awards-season hits. The collaboration aims to explore how AI can assist in creative storytelling and film production — a notable move bringing cutting-edge AI research into Hollywood's creative process.

Microsoft is merging its consumer and enterprise Copilot apps into a single app launching in August. Underused features like Copilot Podcasts are being cut to focus resources. New AI agents called 'AutoPilot' will handle tasks in the background for users who pay extra. This marks Microsoft's push to compete directly with Anthropic and OpenAI in the race to build AI-powered 'super apps' that handle multiple tasks in one place.

The UK's AI Security Institute found that standard AI benchmarks underestimate what AI agents can actually do. When given more compute time (ten times more tokens), AI success rates on software engineering tasks jumped about 25 percent. This means current tests don't reflect real-world AI progress—the actual capability frontier may be around 60 percent steeper than previously thought. It raises questions about how we measure AI safety and readiness.

Google DeepMind employees pushing for union recognition say negotiations with management have stalled just weeks after they began. Workers claim executives aren't engaging in good faith—union representatives say no senior DeepMind leadership attended an initial arbitration meeting, and employees allege Google has restricted internal discussions about the union drive. This marks one of the most significant labor organizing efforts at a major AI lab, triggered partly by Google's 2025 decision to drop its pledge against using AI for weapons and surveillance.

AI-powered bug-hunting tools are generating a flood of vulnerability reports. In June 2026, organizations reported about 1,500 high-severity and critical security flaws—over 3.5 times the previous monthly record. While this means more bugs get fixed, it also means security teams are overwhelmed trying to process, triage, and patch them all. The surge highlights both AI's power for good and the operational strain it creates.